View Full Version : My 4DoF Quad Mech: Numa

06-18-2011, 06:36 PM
Howdy folks,

Some of you know me from the Trossen Chat. I've been part of SocietyOfRobots.com for years. I'm an RL buddy of cire :p

I'm working on my own quadruped, eventually for Mech Warfare. This quad is named Numa. Numa was the second of the ancient Roman kings... and so I can use royalty as an excuse to make a purple bot (e.g. I will eventually dye the brackets purple). And yes, I'm aware of the fat-guy video for Ozone's song, Dragostea Din Tei. I don't mind if you choose to associate the name with that :)

Hardware specs of Numa:
-4DOF per leg
-16 AX-12a servos in the legs
-Axon MCU
-Tristate buffer allows Axon to control the dynamixel servos. (Has purple LED to indicate that servos have power!)
-Can use either Xbee 1.25 mW communication, or serial port communication (Using the Parallax SIP xbee adapter on the robot side)
-11.1 V LiPo battery for servos, 9V NiMh battery for Axon. TBD how I will power camera and guns.
-Using dremel, power switches for Axon power, LiPo power, and DC power supply power are integrated into standard Dynamixel brackets

Obviously I'm shooting for a very similar robot to Immortal, but figuring it out with my own hardware configuration.

Software approach:
-Axon runs my own code created with the modules/etc. from Webbotlib
-Webbot's GaitDesigner format is used for gaits
-I generate my gaits in this format using Mathematica to do the IK calculations. Using its visualization tools is awesome for verifying what my code actually does.
-I will be using VB.Net (Unless I learn a new GUI making tool in the near future) to make my own user interface. Input will quite probably use one of these (http://www.3dconnexion.com/products/spacepilot-pro.html) for awesome points.
-I may eventually have the IK be calculated on the axon, but for now I'm still familiarizing myself with the best ways to do this.

-Beginning of May: Started ordering parts.
-Mid May: Made tristate buffer and was finally able to control servos
-June 12: Finally got the bugs out of my Mathematica code and correctly have my robot walking forwards/backwards
-June 17: successful remote control of walking with xbee link.

Stuff to do/determine
-Give the feet more traction
-How I will move camera/etc. (e.g. I'd save weight by using hobby servos for pan/tilt)
-Optimize the gaits for speed and balance
-Make compensating for changes in center of mass easier
-How to tie the gaits together to do more complicated walking
-Design custom body to nicely integrate all of the parts
-Order the usual Trendnet wireless camera


And now some pictures!:
This pic is a few weeks old, but the body is essentially the same. Switches and accessories have been swapped/added in the mean time.

Animation of top down view of forward walking via Mathematica

Animation of leg movements via Mathematica

The servos' battery and its power connector. I like epoxy.

Combo mutant mount: Axon switch, serial connection interface, and xbee on a parallax adapter.

Axon and tristate buffer are mounted together on a spring-loaded hinged plastic container. The container was the box for a ring.

Another pic of same:

9V battery holder is masking tape + velcro. Real chassis will be nice to have.

06-18-2011, 07:54 PM
And youtube vid.
YouTube - ‪Sloppy Walking‬‏

The transition between poses is what makes it look so choppy. I might move to a more upright gait. I simultaneously need better traction on the wood floor (apartment below me must hate me if they can hear the stomping), and looser traction on the carpet (it fails to go to it's standing up pose). Wheels parallel to the leg direction might be the way to go for that. Obviously balance isn't managed well, currently.

(edit: ok that was confusing, but I fixed the link)

Update: Fixed the turning gait so it's much cleaner. I also bolted these (http://www.trossenrobotics.com/images/gpimages/BPF-F6-150x125.gif) to the back two feet, and that fixes a lot of the balance problems.

06-19-2011, 02:40 AM

:P welcome to mechwarfare

I have to say awfully good start! Love that walking .gif!

06-19-2011, 10:38 AM
Looking good. Keep up the good work..

06-26-2011, 11:43 PM
New youtube vid up. This one has (obvious) corrections to the gaits so that turning is much smoother... other than that nothing new to see there.


-I changed to the smaller sparkfun Xbee explorer.
-Modeling the bot in Solidworks... chassis plates are the main focus for now... The turret gets in the way of things...
-I figured out how I'll do the math for omnidirectional walking... Next I plan to implement this in mathematica, get it cleaned up nicely, then do the code again in C.
-Ordered servos for turret + spares, as well as Lipo battery monitor... Need to determine how much walking time I have with the current setup.

07-12-2011, 11:06 PM
Recent progress took two forms:
-Got battery monitor, and Numa walked ran for 13 mins without the blue LED beginning to flash... Voltage after the fact was 11.7V, iirc... Looks like I should be fine even with more weight once Numa is finished...
-Redid the layout of the 1.25"x1.75" perf-board TriState Buffer. Now it can control either 3 pin or 4 pin servos by closing either of two connections with a jumper pin. I did the design in Inkscape. This allows me to copy/paste all needed documentation into the document for reference. Design and what the screen looks like with this approach are attached.

On the circuit, blue lines are ground wiring, red is 11.1 V battery Vcc, pink is 5 V regulated via Axon, and dashed lines are connections under the board... Short grey lines are solder connections between holes. The 3 header pins at a right angle in the center of the board are where the jumper is placed.

I'll pick up my Trendnet camera this weekend (shipped it to my parents place while NM was on fire.. but it was $40!). Still need to get around to finishing a first iteration (aka allow myself to not have a perfect design), of the so-called chassis plates. Taking the bot apart to tighten screws isn't fun...

07-13-2011, 03:16 AM
Nice work keep it coming

07-13-2011, 08:09 PM
Nice work :)

07-26-2011, 11:30 PM
Some updates!

I continued working on designing that board. I don't know what to call it, but it does a lot of things :P
-Connect 5 chains of Dynamixels (3 or 4 pin)
-Communicate with 3 pin servos (Buffer IC + Hex Inverter IC)
OR (mode selectable with a jumper)
-Communicate with 4 pin servos (Max485 IC)
-Lipo battery monitor connection
-5V regulated for Trendnet camera
-LED indicators for LiPo and 5V (to ICs)

I then built it, of course. Checking numbers, it uses $19 in parts. I guess I see how "simple" boards get "expensive" easily. Updated design and pics of top and bottom attached. The purple surface mount LEDs are on the bottom of the board because I plan to bolt the board to the underside of Numa. The "top" of the board, with all the wires and connectors will be hidden inside the bot, but the LEDs on the bottom will still be easily viewable (also lighting up the ground under Numa = FTW)

The board images are attached at the end. I'm pretty happy with how densely I packed all of those features in there.

I also got the Trendnet camera. I attached it temporarily with zip ties and a bioloid bracket (the cooling slits on the underside are perfect for a zip tie!)

I've got the Trendnet camera working with a Ad-Hoc network, so I should be good for competition where a router is provided...

Bot pics with the camera mounted:

I've also got the chassis plates designed. An initial version will be drilled/cut by my gf since she has access to a machine shop :P. The final version(s) will be carbon fiber. I've got the carbon fiber already, and based on its density of 1.54 g/cm^3, the two chassis plates will weigh less than 50 g combined.

Below is a top down view of my work in CAD so far. The top and bottom plates are duplicated on the right... The holes on the edges of the top plate are for easy mounting of bioloid brackets.


I'm hoping to be able to switch to the new Axon Mote, which is much smaller and would easily fit inside the chassis plates (I've got mounting holes for it, even). It remains to be determined whether I can use the built-in zigbee radio in a similar manner to how I'm using xbees currently.

And one last note: aesthetics!
I've settled on having external wires (e.g. camera power cable) being sheathed in flexible mesh, with shrinkwrap tubing over the ends of the mesh... see the last attached file.

07-27-2011, 10:32 AM
Really digging the home-brew look of it so far. Nice work!

Meccano Jay
07-27-2011, 07:40 PM
Very nice work! The "made not bought" look is a nice touch. Love to see true craftsmanship melding into projects like this.

07-27-2011, 10:13 PM
Thanks for the complements, guys!

Turns out I had the two wires to the Hex inverter (rightmost IC) backwards, and I wasn't getting signals back from the servos. Swapped the wires, and works fine now - nothing fried ;). Also updated the pic in the prev post... and some-day a full schematic.

07-29-2011, 09:53 PM
Today was awesomely productive. I made the carbon fiber chassis plates. All I needed was my dremel :)

Alas, a few holes were in the wrong spots, but I can slot holes really easily... This was due to incorrect dimensions in my model. Every hole I made with the dremel was dead center, thank-ye much :) Fixed those holes in the model for future versions...

The chassis plates + mounting hardware for 5 servos weighs an amazingly low 45.2 g. Cire's Immortal uses aluminum chassis plates that weigh more than double this ;). I'm using 1/32" thick carbon fiber from Mcmaster ($70 for 2 square feet). I've got enough carbon for 4 sets of chassis plates.

I started out by printing the layout on paper, then cutting out the form with my knife (and the pictured cardboard as the surface... I then used packing tape to tape the paper down to the carbon. Packing tape is my personal fave for temporary stuff... doesn't leave any residue. And it's see-through, which is nice for this work.

I then start with the holes, and then do the cutting. Each hole takes about 4 seconds or so with a pointy diamond grit covered bit. The cutting was done with a 3/4" diamond coated dremel disk. Both of these are cheap - $10 for a dozen different bits, and $10 for 11 cut-off discs (three sizes). Oh and of course the dremel, a corded one. Battery powered dremels are dumb...

Here's the bottom plate finished:

And both plates done. Notice the holes along the edges of the top plate... these are spaced for mounting Dynamixel brackets for when I need to test-mount additional stuff...

Bot's rigth side.

Pan and tilt servos mounted perfectly as well

This is where the regular Axon would go. The Axon mote has mounting holes inside the plates...

The underside. You can see where the Axon-Dynamixel interface board goes. The wires go into a hole adjacent to it, and then plug right in. And this is why the indicator LEDs are on the bottom of the board :)

07-30-2011, 04:23 PM
Played with some scraps and made an enclosure for the main power connector and switch. It fits between the two chassis plates, and should be reusable for future versions of the chassis. And it looks nice :D

The enclosure is filled with foam + epoxy, and this adds 3.2g or so...

07-30-2011, 10:02 PM
Wow, looks like its really coming along now.

08-04-2011, 07:41 AM
Looking good! I like the custom made parts, very well done. :)

08-09-2011, 10:57 PM
Thanks for the comments and compliments so far, gents :)

Various updates:

Numa is now re-assembled with the carbon chassis, and walks just as well without modification to the walking gait (omnidirectional walking .... eventually). Showed him off to my parents, and my mentor and his kids/family.

This initial carbon chassis proves that the carbon is plenty strong for my needs (and so easy to work with). I've got the next iteration of chassis plates all CAD'd up, and my GF is working on those in the near future. These next ones and subsequent iterations will be black 3/32" delrin plastic. Once Numa's super finalized, he'll return to carbon.

Got my nylon fasteners and spacers, so now Axon is properly mounted.

I've got the purple dye I'll use on the brackets and nylon fasteners... No idea when I'll get around to that work, though.

Replaced the TV-IP110W with the TV-IP110WN trendnet camera. This one pairs with my wifi card's fake router, unlike the old one (and G devices in general don't seem to work).

Block of wood and four chopped up pieces from broken carbon fiber arrow shafts make Numa's stand for when he's not powered on.

Inverted the camera mounting approach...

And finally, some pics attached :)... More vid eventually...

08-10-2011, 03:40 PM
Owwww, carbon fiber :)

08-10-2011, 11:08 PM
Owwww, carbon fiber :)
Zero carbon fiber ouchies so far. Probably means I know what I'm doing, hehehe.

09-03-2011, 06:20 PM
Some progress...

Of course, I visited Trossen. The walking footage was take two. The first time, one of the screws fell out of a coax joint and walking was sloppy. I had reassembled Numa a couple days before and probably forgot to tighten it then.

Talking with cire, I learned that I'd probably need a thrust bearing for the pan servo. I made that this morning, after CAD'ing up the design a couple weeks ago. I used two layers of 3/32" delrin which I already had... I got lucky with this being the perfect thickness (well sorta). See pics below. This is also a simple design that others could use with AX-12s, so I've attached the PDF... I used the same process as with carbon fiber (print on paper, cut out paper, tape to material, drill/cut with Dremel). One interesting observation: since I'm using delrin (teflon deriv), there's a high melting temperature, so I did not have problems with melted plastic gunking up the dremel bits... Worth the slight premium in price, methinks.

Andrew strongly recommended I cover my exposed Axon, and after seeing Darkback's cracked acrylic faceplate from last year, I'm heeding that advice. ... Attempt one is also pictured below... Earlier this year I found the plastic cover from a ceiling camera in my apartment complex on the floor. Kicked it out of view and took it with me ;). After making the thrust bearing, I spent another 45 mins dremeling the cutout in the plastic cover. Not sure if it's acrylic (bad) or polycarb (good). I'll shoot at it once I get functioning guns. If it's good, I'll either hinge it or otherwise make it easy to open. Cutout is intended for a front hinge, and allows for the Pan servo to do 45° in each direction. Notch is so a connector on the Axon can be unplugged.

Thrust bearing + plastic dome weigh a tad more than the two carbon chassis plates, by the way.... (50.3g) Oh well.

On the topic of guns, got two AEG gearboxes and the new Pololu motor driver that Trossen is stocking. Need to figure out Hop-ups and gun-mounting currently.

Omnidirectional walking math is coded though not tested or put into functions the way I want it yet. Upgrading to Webbotlib 2.x broke compiling for a lot of the test programs I had, so that's going to take time, too.

09-03-2011, 07:47 PM
I can guarantee Giger (and other AEG equipped bots) will shatter the crap out of that globe, haha. It's almost definitely acrylic.

09-04-2011, 09:20 AM
Keep it on then... :tongue:

But make sure you have a back up then.

09-04-2011, 10:48 AM
Well I'm going to shoot at it first :P (AEG gearboxes like I said)

But first I need hopups. Gdubb? :D :D :D

09-04-2011, 11:21 AM
Nice work there "X".. Just hit the dome with a hammer, and see if it breaks..:o Of course it should be at least a 4lbs sledge, and the dome should be attached to the robot to also test the strength and integrity of the various mechanisms, mounts, servos, and carbon fiber.

If all survives, you are ready for guns....

As for hopups... uhhh..lets see....


09-19-2011, 09:49 PM
Quick update. Put Numa's new chassis together this weekend while at Cire's place... Gotta take off the top plate and do more carbon fiber work though. (My awesome girlfriend did most of the dremeling this time while I watched an epic football game; only downside there is that she's not as proficient as I am with the dremel... but all the holes are good!)

Also while at Cire's place, we shot the plastic dome with his airsoft pistol, which he believe to be more powerful than anything on the bots currently... and no damage. So I went ahead and used his Cupcake MakerBot to make a mount to attach the Xbee to it. Also made a few other parts. Those need some dremel work before they're finished, and I'll show those off when I get the chance.

We also tested a modified version of Elaughlin's AEG gearbox hopup with a defender barrel, and it shoots stuff. :) More work to be done on that, but I'm depending on it to be ready for iHobby.

09-19-2011, 10:22 PM
If you get a chance post the CAD file of the hopup...

09-20-2011, 06:51 AM
I just used your file DB, and then made it fit with using the defender barrels and connection pieces. But Ill post it tonight when I get home from work.

09-20-2011, 08:37 AM
It needs modifications for Cire's makerbot to do it with proper tolerances. That'll get posted next weekend when I get a chance to update the CAD a bit...

09-21-2011, 11:26 PM
It needs modifications for Cire's makerbot to do it with proper tolerances. That'll get posted next weekend when I get a chance to update the CAD a bit...
We had to widen everything .020" because it prints everything kind of big, and i took out a lot of the features so that i could print it without overhangs - easier to drill them out afterward anyways. Final version will probably just be a block that fits onto the end of the gearbox, then you drill out everything else by hand so they have tight fits.

09-21-2011, 11:31 PM
We had to widen everything .020" because it prints everything kind of big, and i took out a lot of the features so that i could print it without overhangs - easier to drill them out afterward anyways. Final version will probably just be a block that fits onto the end of the gearbox, then you drill out everything else by hand so they have tight fits.

I suspect, though, that we can do the holes close enough to the right size such that you can ream the holes with a drill bit by hand (e.g. without powering on a drillbit...) Worked for the holes I did with the battery caps.

09-24-2011, 11:50 PM
I mentioned other parts. These were tiny brackets for a hinged clip mechanism for holding sideplates onto Numa. The sideplates will protect various electronics and securely hold the 9V battery for the Axon

I 3d printed some of the parts (yellow) and cut out the other from delrin because they were too small to print nicely. I chopped up the springs from a couple of pens, as well. These contraptions add maybe 20g including the carbon side plates. These things are small. The top yellow pieces have a base of like 1x2cm.

Additionally, a pic showing the inside of Numa... completely packed; and a pic of my current turret design as of earlier today. It's all stuff I can do with the dremel using delrin. (4 weeks to iHobby!)

10-08-2011, 09:55 PM
There's a bunch of pics of the turret at http://www.flickr.com/gertlex

And today, printed and fitted hopups (they work!). Besides coding, the only thing left to do for iHobby (in two weeks) is finish the left hopper and secure the hopups in place...


And Numa doing pushups... turret range is kind a limited and the guns are pretty high up, so this might be helpful against bots that get really close to Numa.

11-04-2011, 07:40 PM
Two weekends ago, I went to iHobby with cire, Immortal, and Numa. Things that went wrong:
-Burned out two AX-12s, a 3rd one probably has a bad motor. (More on this later)
-One AEG motor killed itself (cire has one that did the same thing at some point in the past)
... but I had spares!
-And then my laptop's charger died (laptop's battery isn't good for 2 mins)... of all the things to die... I was able to borrow another guy's laptop and use X-Ctu + IE for control+vid.
-Numa is very tall, and doesn't have as much tilt range in the turret as is needed for bots with very low mounted score panels
-AEG/hopper setup had lots of misfires; working on fixing this.

Things that went well:
-Party at Trossen on the Friday evening
-Got Numa's IK walking working properly at said party (like at 1am)
-Numa walked pretty well
-Won one match out of 6 against Immortal... cause immortal had a leg servo die ;)
-Shot at kids because they asked me to (http://www.youtube.com/user/moverstreet#p/u/2/Tki4DGF4Mn4) (arena wall between them)
-Bought a spare lipo at the show, as well as a 2S 380 mAh battery to replace the NiMH batteries I've been using to power my microcontroller.

Major thanks to Mike and Tim for doing the arena and being good chums. Thanks also, Mike for all the video footage =)
Here's a good one of Numa's walking, as implied above, this is pretty much the IK walking without more than a handful of initial tweaks...:
(This is also when the 3rd servo died on me)


The first AX-12 died while I was working on getting the IK right. I had a bug in my foot-height code that meant the legs were basically slamming into the ground when switching which pair of legs is on the ground whilst walking. The AX with the probable bad motor was found at the same time - it moved as told, but i could move it by hand when it was powered, too.

I'm focusing a bit more on grad school currently, but I aim to redesign the turret for more range of motion and better feeding of BBs. One quick mod I'm playing with is a piece of copper wire going into the hopper to act as an agitator. It's attached to one of the external springs on the gearbox:

11-05-2011, 04:58 PM
Nice fix with the spring, i might try to use :p

12-29-2011, 08:52 AM
So two months of progress for you guys... but really not that much. Grad school, alas.

Physical additions to Numa include:
-second battery monitor for 2S lipo (looks nice having two of them on the front of Numa)
-two Sharp IR rangefinders mounted to the sides (gimmicky proximity sensors that will let me see if I'm too close to a wall...)
-used up all of the wire sheathing that I had to tidy up the wiring
-Changed the feet to use different brackets -> easier walking on carpet and better balance
-Working on getting a whole new set of carbon fiber parts made some of the leg joints will be carbon fiber
-Tested yellow dye for bioloid brackets... gotta get a better yellow (namely Neon Yellow RIT dye, rather than Golden Yellow RIT dye)

I've still got to do the turret redesign. And I burned out another servo's motor, but I'll blame the carpet for that.

And software... lots of work with the command/control stuff. The big thing is that I can now control Numa with an Arbotix Commander... and I'm not using an Arbotix :D. I had to make some fairly straight-forward changes to the Commander.c and .h files so that they work with Axon/Webbotlib variable names. Props to fergs for the Commander communication protocol.

I've also somewhat smoothed the speed changes, but still a bit jerky - not sure if it's a bug, or a limitation of the methods I use. The method works fine in Mathematica! (Basically I track time for the IK with a triangular wave... if I change the frequency (aka Numa's walking speed) of that wave/function, I have to account for two speeds being out of phase, and adjust accordingly)

01-20-2012, 10:22 PM
Some updates!

I successfully did the dyeing, after all (details are in the bioloid dyeing tutorial). Result so far is... so awesome:
(I've gotta dye the feet brackets, but I'm trading grey brackets for white brackets with cire, tomorrow)

The middle leg joints will be replace with carbon fiber... I put my order in this week with Big Blue Saw (leg brackets, new chassis and side plates, whoo!). Downside is, not cheap, and they're not even doing all the 2mm holes in carbon fiber for me (wtf). I'll do the holes myself, fairly painlessly, with aluminum templates that are also getting cut... I was lazy and didn't shop around.

Got my plane ticket to Robogames, too. (It's been an expensive month.)

I redid my turret linkages. Instead of about 40° of tilt range, there's now up to 70° of tilt possible. I've also got nylon spacers that are now dyed. Some of them also have holes and set screws in them. Bunch of pics:

(In case you wondered, the long shiny rods are how I mount the AEG gearboxes)

I can also now control Numa with the Arbotix Commander plugged into my computer, and an xBee plugged into my computer. This allows Numa to send data back to the computer, and the computer forwards the Arbotix Commander signals to Numa. If I want more portability, I just put the xBee on the Commander, and I can do stuff sans computer.

What data does Numa send back? For now it's just two Sharp IR rangefinders that are mounted on the sides. These determine the color of two sidebars next to my video feed... And I've adapted Upgrayd's python based Trendnet viewer for this purpose. The python program also does the forwarding of the Arbotix Commander commands through to the xBee; this does induce some lag, but I may be able to fix this with threads or just running two separate programs.

I've got a vid below that demos the IR rangefinders, Trendnet viewer with colored sidebars, and a bit of walking.


Next up, I need to redo the hoppers so they're actually attached to the gearboxes. The turret redesign saved about 15 g... redoing the hoppers should save another 30-40 g (those sideplates on the turret are heavy). The weight saving is mostly just a secondary challenge, but I like to keep track of it :). I don't use carbon fiber on the turret because CF can block wifi signals...

There's also plenty of programming work to be done... mostly on the IK, which is not as smooth as I'd like; it might just be I need to spend lots of time calibrating so walking is better.

01-21-2012, 10:18 AM
Looking good...The rangefinder info is cool.. not too sure how much they will help, but cool..:cool:


01-22-2012, 10:27 AM
Looking good...The rangefinder info is cool.. not too sure how much they will help, but cool..:cool:


I implemented them as a gimmick sort of thing - in a mech, one has flashing warning lights everywhere, right? :D I also had problems walking too close to walls at iHobby, but I suspect the non-funky buildings at robogames will give a better sense of side-to-side position in a 'street'.

02-24-2012, 10:57 PM
So I've reassembled Numa now that I've dyed everything... even dyed the trendnet camera case, which shrank noticeably.

Oh right, and I had a bunch of carbon fiber parts waterjetted - new chassis plates, leg brackets, chassis side plates, and Arbotix coverplate (for the heck of it). The main cuts were waterjetted, while I did all of the 1/8" and 2mm holes myself. Drilling holes with dremel doesn't shoot carbon dust everywhere, so the waterjetting was nice for my in-apartment fabrication...

I 3D printed 3-part pieces for the feet that span one of the bracket types.

Still gotta fix another bug in the IK.... :( Also, yes, there is a foot servo that is not plugged in below, whoops!

Hopper redesign is a work in progress. My biggest concern is reliable feeding into the AEGs. I hope to sheath the rest of the wires on Numa, but that's the most tedious thing ever. Scoring system is going to be a hat that's easy to remove from the turret, and won't look amazing.


2 Months to Robogames. Oh geeze. In addition to everything else, I need to get a Pelican case and a suitcase...

05-06-2012, 01:25 PM
Well! I successfully completed Numa at Robogames ;) Which is to say, I had everything done except mounting the lighting board when I got to Robogames. Numa was nice and reliable, and got 3rd, despite only scoring 9 hitpoints of damage or so. :P

I ended up redoing my hoppers completely, with reasonable improvement. I did the first gun about 3 weeks before Robogames; the second one a week later, but was a bit more prone to jamming. I basically chopped up two Crosman guns for the 'hopper rails', and then built polycarb hoppers onto those. The rails feed into the 3D printed hopups.

I also added an 'unjammer' for both guns, actuated by a single HS-55 servo (driven by hardware PWM on the Axon). I suspect that having it cycle back and forth (see youtube video) multiple times was a poor choice in that it probably packs the BBs in after the initial poking.

Zero AX-12A motors burned out, amazingly. (I have 5 that have burned out motors…) I think this was mainly due to my adding a crouch mode where torque is disabled in the joint closest to the body. Additionally I made the foot height transitions a bit “smoother”. Finally, I stopped sending the stand pose continuously since someone suggested this might be a culprit of burned motors (still not sure I believe this). But ya, did what I could to avoid burning servos, and it paid off I guess.

Crouch mode did inadvertently lead to Numa tipping during exhibition once: Pushed backwards while in this pose, the back legs curled under the body. When I tried to stand again, the legs were stuck under the body, so the rest of the robot rotated, instead of the back legs. Just something to work around :)

Various other tweaks and things learned over the past two months
-I did sheath all the wires. I also had dyed some of the 3pin connectors yellow, and was replacing connectors as I went. The connectors shrink a tiny bit, enough that the connections can be unreliable.
-I had plugged in the signal wire for the tri-state buffer wrong when I reassembled the chassis a couple months ago... I discovered this two weeks before Robogames. It caused very unreliable communication with the servos for obvious reasons...
-Series 2 (aka ZB) Xbees are to be avoided. Thankfully I had a spare set of Series 1 Xbees. The former are slow to connect and will occasionally drop out.
-Aiming is important… lasers definitely helped other bots, and (at least with Trendnets) are easy enough to see. Definitely adding the lasers that Trossen sells for next year.
-I replaced the 9V battery with a 2S Lipo, which powers the Axon and scoring system. Correspondingly I have a second battery monitor - you can see both of them as the blue LEDs on Numa's front side in the previous post.
-Somehow my 2S battery monitor's blue LED no longer works (broke at RG). Haven't investigated this yet... but it still beeps when I turn the battery on.

Numa will compete next year too… I’ll tweak/improve various stuff, and probably redo the turret/guns/camera again.

Hopper unjammer: (sound is horrendous)


08-01-2012, 07:54 PM
More robogame pics; thanks Tyberius.

And one via a friend.

08-12-2012, 10:58 PM
Printed new feet for Numa over a month ago. Finally got around to putting them on. How this affects walking remains to be seen.

New feet style on the left. Old on the right. The original style were chosen on a whim, rather than thinking of the 'better' new approach.